Automatic speech recognition (ASR) is used to interpret speech received from people in a wide variety of applications. Generally, a person listens to audible prompts or commands and provides spoken responses to those prompts. An ASR system can process the spoken responses to interpret their content and carryout actions that are appropriate. The success with which the ASR system determines the content of the spoken responses can be largely based on the clarity of the spoken responses. However, many people have difficulty clearly enunciating words and phrases. For instance, people may stutter, stammer, or otherwise be unable to generate spoken responses in a timely manner. The ASR system may be programmed using assumptions of what speech sounds like without these difficulties. And as a result, the ASR system may not accurately interpret speech when provided by people who have difficulty speaking.